• Ms. Nainsi Jain
• Mr. Rajneesh Pachouri
• Mr. Anurag Jain

Knee osteoarthritis (KOA) is a progressive joint disorder that significantly affects mobility and quality of life, especially among the elderly. Early and accurate detection is vital for effective treatment planning and prevention of severe joint damage. This study proposes a Region of Interest (ROI)-focused deep learning approach for the automated detection and classification of KOA severity from knee X-ray images. The system employs two convolutional neural network models, MobileNetV2 and VGG16, chosen for their efficiency and high classification performance. A novel preprocessing technique based on pixel density is introduced to automatically extract the cartilage region from X-ray images, enhancing feature relevance and reducing background noise. The dataset consists of 1,650 high-quality grayscale X-ray images, manually annotated by medical experts using the Kellgren and Lawrence grading system. The MobileNetV2 model achieved a test accuracy of 96%, outperforming VGG16, which attained 92%. The system is implemented with a Flask-based web interface for real-time usability, offering a scalable and accessible solution for clinical deployment. Comprehensive evaluation metrics, including accuracy, precision, recall, and F1-score, confirm the system’s effectiveness in automating KOA assessment, particularly in resource-constrained healthcare environments.

Knee Osteoarthritis, Deep Learning, MobileNetV2, VGG16, X-ray Imaging, Region of Interest (ROI).

"ROI-Focused KOA Detection Using MobileNetV2 and VGG16", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 9, page no.f498-f504, September-2025, Available :http://www.jetir.org/papers/JETIR2509565.pdf

"ROI-Focused KOA Detection Using MobileNetV2 and VGG16", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 9, page no. ppf498-f504, September-2025, Available at : http://www.jetir.org/papers/JETIR2509565.pdf